Posttransplant De Novo Donor-Specific HLA Antibodies Identify Pediatric Kidney Recipients at Risk for Late Antibody-Mediated Rejection


  • Correction made after online publication September 7, 2012: author affiliations have been updated.

Fabrizio Ginevri,


The emerging role of humoral immunity in the pathogenesis of chronic allograft damage has prompted research aimed at assessing the role of anti-HLA antibody (Ab) monitoring as a tool to predict allograft outcome. Data on the natural history of allografts in children developing de novo Ab after transplantation are limited. Utilizing sera collected pretransplant, and serially posttransplant, we retrospectively evaluated 82 consecutive primary pediatric kidney recipients, without pretransplant donor-specific antibodies (DSA), for de novo Ab occurrence, and compared results with clinical–pathologic data. At 4.3-year follow up, 19 patients (23%) developed de novo DSA whereas 24 had de novo non-DSA (NDSA, 29%). DSA appeared at a median time of 24 months after transplantation and were mostly directed to HLA-DQ antigens. Among the 82 patients, eight developed late/chronic active C4d+ antibody-mediated rejection (AMR), and four C4d-negative AMR. Late AMR correlated with DSA (p < 0.01), whose development preceded AMR by 1-year median time. Patients with DSA had a median serum creatinine of 1.44 mg/dL at follow up, significantly higher than NDSA and Ab-negative patients (p < 0.005). In our pediatric cohort, DSA identify patients at risk of renal dysfunction, AMR and graft loss; treatment started at Ab emergence might prevent AMR occurrence and/or progression to graft failure.




antibody-mediated rejection


complement dependent cytotoxicity


cross-reactive epitope groups


cyclosporine A


delayed graft function


donor-specific antibodies


mean fluorescence intensity


nondonor-specific antibodies


Although in recent years short-term renal allograft outcome has significantly and progressively improved, late failure of kidney transplants remains a relevant clinical problem (1). Thus, clinical research is focusing toward a better understanding of the mechanisms leading to chronic graft damage (2–4), to identify biomarkers able to predict its onset and guide early or preemptive intervention.

The detrimental role of HLA antibodies on allograft outcome has been extensively evaluated (5,6). The recent evolution of alloantibody detection assays from cell-based to solid-phase methods (7) has facilitated studies that demonstrated a positive association between presence of HLA antibodies before/after transplantation and poor transplant outcome, and significantly contributed to elucidate the central role of humoral immunity in the pathogenesis of chronic allograft damage (2,8–13). The majority of these studies evaluated the role of HLA antibodies in cohorts partly, or exclusively, including sensitized patients, with analysis conducted within the first posttransplant year, or at a single point in the posttransplant follow up.

Among the pretransplant solid-phase immunoassay-negative, low-risk group, a number of patients will develop chronic allograft dysfunction and, ultimately, graft loss. A recent study demonstrated that, within this cohort, appearance of de novo donor-specific HLA antibodies (DSA) preceded development of antibody-mediated injury, that led to a 40% decrease in 10-year graft survival (14). These results suggest that surveillance for posttransplant HLA antibodies could predict graft dysfunction and have an impact on therapeutic management and clinical outcome even in these low immunologic risk patients.

As evidence in the pediatric cohort, which includes a large percentage of HLA antibody-negative patients at the time of first transplant, is so far very limited and controversial (15–17), we analyzed a pediatric cohort of pretransplant HLA-antibody-negative, first kidney recipients for: (i) de novo occurrence of anti-HLA antibodies through a retrospective, systematic evaluation of sequential serum samples; (ii) HLA antigen specificity and function characterization; (iii) relationship of anti-HLA antibody emergence with renal function and graft outcome.

Patients and Methods


Ninety-one consecutive pediatric patients, referred between March 2003 and December 2010 to the Genoa Pediatric Kidney Transplant Program for first allografting, were included in this study. All grafts were performed after a negative T cell cross-match carried out by complement dependent cytotoxicity standard method.

Our standard of care for low immunological risk kidney transplant patients consisted of induction with basiliximab, and a triple drug immunosuppressive regimen including a calcineurin inhibitor, mycophenolate mofetil and prednisone. Biopsy-proven acute cellular rejection (ACR) was treated with pulse intravenous methylprednisolone.

Graft biopsies were performed only for clinical indication. Rejections were histologically graded following Banff 97 criteria with updates. Banff 09 criteria were employed for classifying C4d positive acute antibody-mediated rejection (AMR; 18). C4d staining was performed on frozen sections by indirect immunofluorescence. C4d negative AMR was considered in donor-specific antibody (DSA)-positive patients with negative C4d staining and any of the following microcirculation lesions: peritubular capillaritis, glomerulitis, thromboses and transplant glomerulopathy (4,19). According to the same criteria, we defined a probable C4d negative AMR in a patient with non-DSA (NDSA).

Experimental design

A unique source of sera, collected before transplantation, 3-monthly in the first posttransplant year, and annually thereafter, was available for all 91 patients. Only patients with 1-year minimum follow up were included in the analysis. Recipients of first graft who were found positive for the presence of anti-HLA antibodies in the pretransplant sera (n = 4), as well as three patients with combined kidney and liver transplantation, and two patients who had a graft loss within six months posttransplant because of focal glomerulosclerosis recurrence, were not included. Demographics of the 82 kidney recipients included in the final analysis are summarized in Table 1. This study was conducted according to Institutional Review Board guidelines.

Table 1.  Demographics of patients analyzed in the study
Recipient age (years, median and range)14 (2–27)
Recipient sex (F/M)46/36
Kidney donor (living/deceased donor)12/70
Donor age (years, median and range)13.5 (1–55)
Donor sex (F/M)51/31
HLA-A,B,DRB1 mismatches (mean ± SD)3.1 ± 1.05
Follow up (years, median and range)4.3 (1–8.7)
Baseline immunosuppression
 Anti-CD25 mAb82
 CNI alone/CNI+MMF8/74
Delayed graft function13
Cause of end-stage renal disease
 Renal hypodysplasia/dysplasia32
 Nephrotic syndrome 
   Focal glomerulosclerosis13
   Finnish type nephrotic syndrome 4
   Diffuse mesangial sclerosis 2
 Alport syndrome 5
 ANCA positive vasculitis 2
 Others 4
 Unknown 4

HLA class I and class II typing

Recipient low-resolution HLA-A, B, DRB1 typing was performed with a microarray bead-based technique (Lambda Array Beads Multi-Analyte System LABMAS, One Lambda, Canoga Park, CA, USA). HLA typing of donors was performed by both serology (GTI, Waukesha, WI, USA) and by PCR–sequence specific primers (ABDR SSP Combi Tray, Olerup SSP, Saltsjöbaden, Sweden). When required, HLA-C, DQB1 and DPB1 typing were performed. To confirm antibody specificity at allelic level, high resolution HLA typing with sequence based typing was also carried out. We used locus specific primers (Atria Genetics, San Francisco, CA, USA). DNA sequences were obtained after processing with 3130 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) and analyzed with Assign 3.5+ software (Conexio Genomics, Applecross, WA, Australia).

HLA antibody screening with microarray bead-based assay

IgG and IgM anti-HLA reactivity in the sera was tested with a bead-based screening assay (7). Briefly, we used the LABScreen Mixed kit (One Lambda) which simultaneously detects class I and class II antibodies with microbeads coated with purified class I and class II HLA antigens. Results above a cut-off value of 3.0 were considered positive. The Single Antigen kit (One Lambda) was also used to identify HLA specificities, including HLA Cw and HLA DQ, DP and DQα antigens; analysis was performed with One Lambda software (HLA Visual Version 2.2).

The serum displaying the highest DSA reactivity from each patient was also tested with C1qScreen™ (One Lambda) for identification of complement binding antibodies (CBA). R-Phycoerythrin labeled anti-human C1q antibody was employed. Fluorescence intensity, measured on a Luminex analyzer, indicates the relative amount of antibody bound to the test sample.

All sera with a Mean Fluorescence Intensity (MFI) value ≥1000 were considered positive in both single antigen and C1q screen.

Statistical analysis

Data were expressed as mean ± SD or as median and range, as appropriate. The correlation of clinical parameters with HLA antibody presence was evaluated by Student's t-test. The contribution of various risk factors to HLA antibody development was evaluated by chi-square test. Comparison of immunological and clinical parameters among different patient subgroups was performed by the Kruskal–Wallis one-way analysis of variance. p < 0.05 were considered statistically significant. Statistical analysis was performed using the NCSS System (NCSS, Cary, NC, USA).


Incidence and characterization of de novo HLA antibodies

Among the 82 patients analyzed, observation time ranged between 1 and 8.7 years, with a median time of 4.3 years.

Thirty-seven patients (45%) developed de novo HLA antibodies: 19 patients had DSA (23%), with six showing positivity for both DSA and NDSA, whereas the other 18 developed NDSA only (24 total patients with NDSA, 29%; Figure 1). According to their antigen specificity we observed:

Figure 1.

Cumulative risk of developing DSA and NDSA in pediatric recipients of kidney transplantation. N = total number of patients; E = events (number of patients developing DSA or NDSA). Cumulative incidence at 10 years and 95% CI are reported.

  • 1Patients with DSA (Tables S1 and S2): Of the 19 patients developing DSA, the two patients mismatched for class I only, developed class I DSA exclusively, as expected. Among the other 17 kidney recipients, all HLA-class I and -class II mismatched with the donor, the majority (n = 11) developed class II DSA only, whereas six displayed both class I- and class II-specific antibodies. Class II antibodies were almost exclusively directed against HLA-DQ specificities.
  • 2Patients with NDSA only (Table S3): The majority of patients in this group (15/18) developed class I- and/or class II-specific NDSA that recognized cross-reactive epitope groups (CREG) specificities related to donor mismatched HLA antigens. Non-CREG-directed antibodies were present in six cases. Among CREG-directed antibodies, the majority (13/15) were class I-specific. Class II antibodies were directed against DR in two cases and DQ in three cases.

We then analyzed de novo antibody MFI values. DSA showed a significantly higher peak MFI than NDSA (median 9500 vs. 2500, p < 0.0005; Figure 2, panel A). Within DSA, class II-specific antibodies showed a trend toward higher median peak levels (12 000 vs. 4550 of class I-specific DSA; p = 0.06). Considering HLA class II-specific antibodies, peak MFI was significantly higher for DSA than NDSA (median levels: 12 000 vs. 1550, respectively; p < 0.001; Figure 2, panel B).

Figure 2.

Peak MFI of anti-HLA antibodies in the different study groups. Panel A: peak MFI of DSA and NDSA. Panel B: peak MFI of anti-HLA class-I and II DSA and NDSA. Results are reported as single peak values (black dots) and medians (black lines). Differences among groups were analyzed by Mann–Whitney test. p < 0.05 are reported in bold; p-values between 0.05 and 0.1 are also reported.

Timing of de novo anti-HLA antibody emergence

DSA appeared at a median time of 24 months after transplantation (range 3–60 months; Figure 3). NDSA developed earlier than DSA (median time: 16.5 months), although the difference was not statistically significant. Within the DSA group, there was a trend toward an earlier development of HLA class II versus class I antibodies (median: 24 months vs. 48 months, p = 0.08).

Figure 3.

Time course of HLA antibody detection and relevant clinical events in the DSA patient group. Serial posttransplant HLA antibody assessments, appearance of NDSA and DSA, for-cause graft biopsies and histology outcome and graft losses are reported for each patient developing DSA. On the Y-axis DSA patient ID is reported; on the X-axis, time after allograft, reported as months or years.

The majority of DSA-positive patients showed prolonged positivity, that persisted during the follow up (Figure 3). In five patients, first DSA positivity was followed by Ab disappearance. In four of these five patients DSA subsequently reappeared. Conversely, NDSA disappeared at follow up in more than 50% of the patients (data not shown).

Correlation of anti-HLA antibodies with clinical parameters

No difference in age at transplant, sex, living donor graft use, number and class of matched donor-recipient HLA alleles, delayed graft function, ACR and development of cytomegalovirus, Epstein–Barr virus and polyoma BK virus replication, existed between patients with or without evidence of HLA antibodies. We observed a significantly increased risk of developing DSA in patients receiving CyA-based versus tacrolimus-based immunosuppressive regimen (16/56 patients, 29% vs. 3/26, 11% in the tacrolimus group, p < 0.05).

Renal function, measured as serum creatinine levels at discharge, at the time of anti-HLA antibody detection in positive patients, and at the end of follow up, was analyzed in the entire cohort.

Patients who did not develop anti HLA antibodies and those showing NDSA had comparable serum creatinine levels at discharge and throughout the follow up (Figure 4, panels A and B). Conversely, a significant increase in creatinine levels was observed in the DSA patient group at the end of follow up, compared with values observed at discharge and at the time of first DSA appearance (median levels of 1.44 mg/dL vs. 1 mg/dL vs. 1mg/dL, respectively; p < 0.005; Figure 4, panel C).

Figure 4.

Renal function in the different study groups. Serum creatinine values (mg/dL) measured at the time of discharge after transplantation, at the time of first anti-HLA antibody detection and at the end of follow up (A) in patients who do not develop anti-HLA antibodies, (B) in the NDSA group and (C) in the DSA group are shown. Results are reported as single values (black dots) and medians (black lines).

At the end of follow up, a significant serum creatinine increase was observed in the DSA group as compared to the control and NDSA groups (1.44 mg/dL vs. 0.9 mg/dL vs. 1.05 mg/dL, respectively; p < 0.0001). This finding was in line with the observation of five graft losses (four due to late AMR, one due to amyloidosis in the grafted kidney) in the DSA group (26%; Figure 3) as compared to none in the NDSA group, and one (due to focal glomerulosclerosis recurrence) in the Ab-negative group (2%).

Correlation of HLA antibodies with pathology findings

Kidney graft biopsies were performed only in the presence of clinical indication. The pathological findings were analyzed in relation to the serological status of each patient (Table 2, Tables S4 and S5). We found that:

Table 2.  Pathology findings according to HLA antibody presence
DSA n = 19NDSA n = 18Negative n = 45
  1. AMR = antibody-mediated rejection; TCMR = T cell mediated rejection.

Number of patients with biopsy17/19 (89%)8/18 (44%)5/45 (11%)<0.0001
AMR11/17 (65%)1/8 (12.5%)0/5 (0%)<0.01
 C4d positive 8/17 (47%) 
 C4d negative 3/17 (18%)1/8 (12.5%) 
 ≥1A 4/17 (24%)0/8 (0%)3/5 (60%)<0.05
  • (i) Seventeen of the 19 patients with DSA (89%) had an indication to undergo kidney biopsy, as compared to 44% of patients belonging to the NDSA group (8/18) and to 11% of HLA antibody-negative patients (5/45; p < 0.0001);
  • (ii) In a large proportion (13/17) of DSA patients, kidney biopsies were performed subsequent to emergence of anti-HLA antibodies (Figure 3), whereas most NDSA patients underwent kidney biopsy before antibody appearance (5/8);
  • (iii) Within the DSA positive group, chronic active AMR was observed in 11 of the 17 biopsied patients (65%). 8/11 patients from this group showed a C4d diffuse staining positivity (Table S4). DSA preceded AMR by a median of 1 year.
  • (iv) None of the patients in the NDSA group experienced a C4d+ AMR; one patient in this group, whose biopsy was performed beyond the first posttransplant year, showed features of microcirculation inflammation compatible with a probable AMR (Table S5). One could argue that, in NDSA group, biopsies were prevalently done before antibody appearance. However, it should be taken into account that in this group no other biopsies were required in the subsequent follow-up period, as renal function was stable.

Taken together, these results indicate that antibody-mediated graft damage is prevalently, if not exclusively, present in patients developing de novo DSA.

At present, the limited size of our cohort does not allow to gain further insight into the respective role of HLA class I and class II Ab, or Ab titer change, in the pathogenesis of tissue injury. Indeed, although HLA class I Ab were present in a smaller percentage of patients, their presence was associated with AMR and graft loss, as observed for HLA class II alone (Figure 3).

We also investigated whether complement binding activity by HLA Abs might have an effect on renal function. Among the 19 DSA-positive patients, donor-specific CBA were found in nine patients (47%). No difference in terms of graft loss, cellular and humoral signs of graft damage and graft dysfunction was found between the patients displaying or not complement-binding DSA (Table 3).

Table 3.  Pathology findings and renal function according to positivity for C1q fixing assay in DSA patients
 Positive C1q (n = 9)Negative C1q (n = 10)
  1. AMR = antibody-mediated rejection.

Graft loss23
 C4d positive44
 C4d negative21
Serum creatinine at discharge (median, range)1 (0.4–1.5)1.05 (0.7–3.2)
Serum creatinine at Ab appearance (median, range)0.9 (0.6–1.4)1.25 (0.8–3.8)
Serum creatinine at end follow up (median, range)1.4 (0.7–7)1.52 (0.7–6.9)


In our study, we conducted a systematic analysis of posttransplant de novo HLA antibody development in a cohort of unsensitized pediatric kidney recipients. This design allowed us to assess in a unbiased fashion the impact of de novo HLA antibodies on clinical outcome, as the allograft natural history was not modified by any therapeutic intervention until the occurrence of relevant clinical events. The main finding of this longitudinal survey, in line with most published studies in adult kidney recipients (2–4,8–14,20,21) but at difference to what reported in a recent pediatric cohort (17), is that the emergence and persistence of de novo DSA, which occurred in as many as 23% of the patients over a median period of 4.3 years, significantly correlated with worsening of renal function, AMR and graft loss. Although we only performed for cause biopsies, it is relevant that, in our cohort, (i) many events requiring a biopsy occurred at/after DSA development, (ii) AMR was exclusively present in the DSA group and (iii) DSA development mostly preceded AMR. Moreover, patients in the DSA group had a significantly worse renal function at follow up than at the time of first DSA detection. In the light of these observations, protocol biopsies would have likely facilitated an early chronic AMR diagnosis, at least in those developing DSA within the first 2 posttransplant years.

In addition to specificity and titer, it has been suggested that HLA antibody functional activity may also have an impact in the induction of graft damage (22). However, as recently shown in heart transplant recipients (23), we found that the presence of persistent DSA has a negative association with graft outcome independently of their capability of complement fixation, although this finding might reflect a bias in the experimental design, as with our choice to test a single time at the highest MFI we may have missed some positivities (24).

When analyzing parameters possibly correlated with HLA antibody development, we did not observe any association with known risk factors, such as HLA antigen mismatches, or previous ACR (14), probably due to the relatively limited size of our pediatric cohort and to the absence of surveillance kidney allograft biopsies. However, in the presence of follow up levels of immunosuppressive drugs within protocol range and comparable in DSA, NDSA and negative groups, we observed a significantly increased risk for DSA development in patients receiving CyA as opposed to tacrolimus-based immunosuppression, suggesting that tacrolimus may be more effective in preventing T cell-mediated antibody formation.

In our cohort, de novo NDSA alone, which were observed in 22% of the patients posttransplant, had no impact on graft function deterioration. These data are in line with two recent studies conducted in adult kidney recipients (14,20), whereas a body of previous evidence indicated that the presence of NDSA was also associated with kidney dysfunction and graft loss (8,9,12,21). Because the tight and prolonged screening schedule support our observation, this apparent discrepancy may be explained by the different patient selection criteria employed. As in the cohort reported by Wiebe et al. (14), our patients have been purposely selected for being at low immunologic risk, whereas the majority of published studies included sensitized patients, or assessed pretransplant antibodies by less sensitive methods, and posttransplant HLA antibody evaluation was often performed on a single sample, or limited to the early posttransplant period. Thus, NDSA may not have been de novo, or patients in the NDSA group may have been misclassified. Alternatively, on the basis of our findings, it is reasonable to hypothesize that, in nonsensitized recipients, the immunological stimulation, initiated by a variety of public epitopes shared with mismatched donor antigens, induces a weak alloresponse to CREG specificities leading to a transient NDSA production (25), devoid of pathologic consequences. In sensitized recipients, or under particular clinical conditions, this early alloresponse could be amplified, causing persistent production of high level NDSA, or facilitating emergence of stronger reactivities to donor-specific HLA or non-HLA antigens, eventually leading to graft damage.

Our study shows that in a low-risk pediatric population on conventional CNI-based triple drug regimen, development of de novo DSA, that occurs in almost a fourth of the patients, but not NDSA, is associated with late/chronic active AMR and poor graft outcome. Procedures such as antibody removal by plasmapheresis or extracorporeal absorption, antibody production downregulation by B cell and/or plasmacell targeting, or complement cascade inhibition, have had a very limited success when employed in an advanced phase of AMR (26–29).

As appearance of DSA may precede AMR and renal dysfunction by years, it seems reasonable to consider DSA development as a predictor of long-term dysfunction and proceed to some form of intervention, to minimize the risk of graft loss (30). Controlled trials will be needed to assess whether an early DSA removal and/or production inhibition, perhaps in association with maintenance immunosuppression implementation, could be successful in preventing and/or delaying chronic graft damage. Despite the presence of renal dysfunction in a sizeable proportion of patients at a medium-term observation, the follow up in our and other cohorts is yet too short to have a clear prediction of the progression rate to poor graft outcome in the DSA group. Thus, it may be argued that available evidence is insufficient to justify a complex treatment strategy not devoid of potentially serious side effects. A better characterization of alloantibodies could help identify, within the DSA cohort, those patients more at risk of graft damage, to further restrict preemptive treatment. Alternatively, because in our as well as in other studies, DSA were prevalently directed to class II antigens, one could envisage to prevent alloantibody formation by a better HLA class II matching during the organ allocation process.


Supported in part by grants from “Cinque per mille IRPEF-Finanziamento della Ricerca Sanitaria Istituto G. Gaslini”, to FG and GMG; Istituto G. Gaslini, progetti Ricerca Corrente, Ministero della Salute (contributo per la ricerca intramurale) to FG and GMG; Progetti Ricerca Finalizzata, Ministero della Salute, to PC and MZ; AIRC (Associazione Italiana Ricerca sul Cancro) to PC; grant from Fondazione La Nuova Speranza Onlus to FG, PC and GMG. FG and MC are recipients of grants from Fondazione Malattie Renali del Bambino.


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.